Axiom Home Page Forums General Discussion Stereo A difference in Vinyl & CD - Seeking explanations?

I know the ancient analogue/vinyl vs. digital/cd topic has been covered endlessly in other forums, but I'm curious about the opinions of people here as to the ongoing benefits of vinyl, and in particular wonder if you can help me account for a specific difference on highs. Last night I did some comparison tests on John Coltrane's Blue Train album, comparing that song, "Blue Train," from the LP that Blue Note reissued in 1990 to the version on "The Last Giant: The John Coltrane Anthology" from Rhino Records. The Rhino CD offers the same take from the original Blue Train album, from the 1957 recording session.

Listening to them both on the M60s, and adjusting for volume as I switched between (since the LP plays at a distinctly lower volume), the differences were extremely apparent. The bass lines were great on both, but the highs, particularly a cymbal-brushing sequence, sounded so different it was like two different recordings. On CD the cymbal brushes leapt out of the right speaker, and were frankly distracting. On vinyl they blended in and seemed to assume their proper place. There were plenty of other subtle differences -- and I enjoyed both formats, with a slight preference for vinyl on this kind of jazz. But the cymbal-brush difference was so pronounced -- do you think the CD mix was just different, or is this part of the sonic difference between the formats?

Birdman

I say yes to all of the above.

The fact that you noticed the highs having the most prominent difference is not surprising. I take it you're familiar with the digitizing process and how a CD contains 44,100 samples per second, and each sample is a 16-bit 'word'.

Now picture two soundwaves, one of an upright bass playing a low note and the other of a brushed cymbal. There's a lot more going on in the brushed symbal soundwave. In addition to being more textured, the sound is much higher in the audible spectrum and thus oscillates many more times per second than the bass note. The more complex a soundwave, the more is potentially lost upon digitization. (IMHO, CD audio is a compressed format since it discards data by design during the mastering process.)

This same phenomenon plagues MP3s. If you ever listen to MP3s, the highs are where you notice most of the compression artifacts.

Hi birdman,

I think it's likely both factors. Vinyl typically has higher frequencies like brushed cymbals somewhat rolled off. On average, highs between 10 kHz and 20 kHz on vinyl are about -3 dB below those of the same tape mastered to CD, except for some of the direct-to-disc and 45-rpm mastered LPs like The Sheffield Drum Record, which first on vinyl and then reissued on CD remains the best recording of a drum kit ever made.
Mastering engineers for vinyl routinely would tweak the high frequencies (roll them off a bit) so cheap phono cartridges would track them without distortion.

By the way, CD is not a "lossy" algorithm like MP3. The Nyquist theorem demonstrates that a digital sampling of a 20-kHz frequency--so long as you use a 44-kHz sampling frequency, which CD does--will perfectly reconstruct the 20-kHz frequency. On instant A/B comparisons of the half-speed mastered vinyl of the Sheffield Drum record vs. the CD reissue, both running in sync., apart from occasionally audible LP surface noise, I'd challenge anyone to be able to tell them apart. (I used to set up comparisons like this when CDs were first introduced in the early '80s.)

Of course, there are huge differences introduced by tweaks in EQ on a CD remaster. Some are really awful; others improve on the original. I have three different vinyl issues of a classic jazz vocal album by Lambert, Hendricks & Ross, one pressed in Italy that's really nice, plus a couple of CD reissues. Some of them are very different in tonal balance, even the vinyl. The best-sounding ones are the Italian vinyl pressing and an American CD digital re-mastering of a few years ago. There was an American vinyl re-issue in the 1980s where the vocals had a lot of midrange treble emphasis that makes the LP all but unlistenable.

Regards,

Interesting- I've always heard the Nyquist theorem referenced to show limitations of digitization (i.e. the maximum frequency that a 44.1khz medium can encode is 44.1khz/2 = 22.05khz- any frequency higher than that will hover near 22khz and distort substantially)

Can the theorem be applied conversely to prove that a perfect reconstruction/encoding of a frequency occurs in a digital medium given the absence of this theoretical impossibility? (would 200 samples/second be just as good/accurate when digitizing a 99.487hz tone as 44.1k samples/second?)

I wish I knew more about this stuff! Makes me wish I hadn't skipped so many math classes...

Very interesting, Alan - thanks for the info. Pls pass along (anyone) any online resources for learning more on this.

Ah, but Peter, that's the magic of FFT, interference, and all that other stuff I don't really remember the name of from Physics/Astronomy class. I do remember, however, that you don't lose much (if anything) from doing it. You were asleep that day, weren't you? ;-)

FFT = Fast Fourier Transforms?

Yup. I'm assuming that those are part of what's used... But I kind of slept through that stuff, too...

Re "Can the theorem be applied conversely to prove that a perfect reconstruction/encoding of a frequency occurs in a digital medium"?

The short answer is yes. See Nyquist Definition

"Nyquist rate is the minimum theoretical sampling rate that fully describes a given band-limited signal, i.e., enables its faithful reconstruction from the samples."

Given that (I assume) redbook CD is capable of achieving this Nyquist rate, is SACD/DVD-A complete overkill?

That's controversial. There may be some benefit in higher sample rates, but how scientifically justifiable it is, I don't know. You can't go by allegations that a high-resolution SACD or DVD-A higher sample rates "sound better", as usually other concurrent changes exist, such as re-mix and digital cleanup from original master tapes, etc. SACD and DVD-A often sound better, but whether this is solely (or even primarily) due to the higher sample rate seems debatable.

For example DVD-A discs usually have a Digital Dolby or DTS 5.1 compatibility track so they'll play on existing DVD players without requiring a specialized DVD-A player. The DD 5.1 max aggregate sample rate for all 6 channels is 448k/sec. Even if you discard the LFE track and assume the total bandwidth is available for the remaining 5 channels, that's only 89k/sec. By comparison SACD sample rate is an incredible 2.8 million/sec. Yet to my ear, the DD 5.1 track on my DVD-A discs (89k samples/sec/channel) sounds equally good as my SACD discs (2.8 million samples/sec).

They're both generally better than regular CDs, but how much is due to the higher sample rate is unclear. That DD 5.1 sounds so good seems to indicate going much higher that 89k/sec is overkill.

For simplicity I omitted that SACD DSD is 1-bit samples vs CD or DD 5.1 using 16 to 24-bit words.

In addition to the theoretical higher sample rate advantages, most current SACD and DVD-A players have disadvantages vs CDs and DD/DTS 5.1: they require 6-channel analog connections to your amp, so your amp's fancy digitally-implemented bass crossover, calibration, EQ, etc is useless. Eventually this will get fixed, but it's possible the very real analog-related disadvantages of most current SACD/DVD-A players may swamp the theoretical advantage of higher sample rate.

Peter, as I said in an earlier reply, my view is similar to Joe's on this point. As set forth by the motion picture music editor's views I linked there, little if any of the improved sound is due to more bits or higher sampling rates, but rather to multi-channel surround, better remixing, better remastering, etc.